life as a post-doc at the national cancer institute€¦ · as a post-doc, my main job is to...

Introduction My projects Post-doc Details From UNL to Post-doc

Life as a Post-doc at the National Cancer Institute

Pavel Chernyavskiy, Ph.D.Alpha Seminar

Nov 21nd, 2016

About me

Born in the Ukraine in 1985, immigrated to New YorkCity in 1996

BS in Economics from SUNY - Binghamton in 2007

MS in Statistics from UNL in 2012; PhD in 2015

Post-doc at the National Cancer Institute (NCI) since July2015

What I do

As a post-doc, my main job is to conduct research andpublish papers

Publishing my dissertation is part of my jobFinding a permanent position is part of my job

My research broadly:

Methods and applications using spatial andspatio-temporal statistics

Incorporate spatial thinking into problems wherespatial thinking has not been used before

What I do

As a post-doc, my main job is to conduct research andpublish papers

Publishing my dissertation is part of my jobFinding a permanent position is part of my job

My research broadly:

Methods and applications using spatial andspatio-temporal statistics

Incorporate spatial thinking into problems wherespatial thinking has not been used before

Dissertation

Spatio-temporal modeling of EEG data collected from college athletesUsed model parameters to di�erentiate between athletes with andwithout traumatic brain injuries (submitted to JASA)

Post-doctoral projects

Spatial interpolation of indoor background gamma radiation inthe United Kingdom (Chernyavskiy et al., 2016; Journal of Env Rad)

Premature mortality in the U.S. by race and sex use standardage-period-cohort models (Shiels et al., 2016; in revision for The Lancet)

% change per year in total mortality for men and women of di�erentages, races (1999-2014)

Developed a more e�cient parameterization of the age-period-cohortmodel (in review, Stat in Med)Used a spatial random e�ect to capture what is left over after age trendand temporal trend were estimated.

+ =

Developed a more e�cient parameterizationof the age-period-cohort model (in review,Stat in Med)

Applied new model to brain cancer and breast

cancer incidence rates (A, C are observed

rates; B,D are predicted rates)

Spatially-varying age-period-cohort model with random e�ects speci�edusing multivariate disease mappingU.S. state-level mortality for white non-Hispanic women aged 25-50

Structure

Duration of post-doc: 3 years; 2 addtl. years may granted

Money matters:

Salary: statisticians start at ~$64,500, increase by set amountsevery yearTravel: $2,000 per year (1 domestic conference)Additional travel funding available through paper competitionsIneligible to apply to most federal grants

Mentorship:

one mentor from the Biostatistics Branch, one mentor fromRadiation Epidemiology Branchpost-docs are free to collaborate, pursue independent projects,mentor undergraduate fellows

All manuscripts, presentations must �rst pass internal review

Structure

Money matters:

Mentorship:

Structure

Money matters:

Mentorship:

Structure

Money matters:

Mentorship:

The Good and the Bad

Favorite things about post-doc:

1 No work hours per se, time o� not tracked very closely

2 Total control over projects I work on

3 Responsive, engaged, accomplished mentors

4 Intelligent, diverse fellows

5 My computer, �Alice�. Alice has 16 cores and 192GB ofmemory and she is all mine ,

Less-than favorite things about post-doc:

1 Not the best preparation for academic career:

1 Most investigators don't publish in Statistics journals2 Pressure to publish quickly; review time in Statistics journals

too long3 Most mentors never held academic positions, may not know

what makes fellows marketable

2 Salary is good for a post-doc; not so good for a Statistics PhD

3 Internal review process can be annoying

4 I don't get to bring Alice with me after post-doc /

What I use the most

Software:

R, LYX (just Beamer)WinBUGS, JAGS, Stan (just starting)EndNote (for reference management)

Statistical skills:

Ability to read & understand papers in Statistics journals,medical journals, etc.Ability to write, verbally communicate with statistical &non-statistical audienceGeneralized Linear Mixed Models (especially the Poisson)Matrix algebra (not integration)Multivariate Normal distribution (because of spatial statistics)

What I use the most

Software:

R, LYX (just Beamer)WinBUGS, JAGS, Stan (just starting)EndNote (for reference management)

Statistical skills:

Ability to read & understand papers in Statistics journals,medical journals, etc.Ability to write, verbally communicate with statistical &non-statistical audienceGeneralized Linear Mixed Models (especially the Poisson)Matrix algebra (not integration)Multivariate Normal distribution (because of spatial statistics)

What I wish I learned

Bayesian statistics

Sparse matrices

E-M algorithm

A �lower-level� programming language like C or Fortran

Genetics

General advice

Graduate school and academic lifestyle:

Learn to fail �better�, learn to �get over� failures quicklyStand up for your research: argue (respectfully) with yourmentors/advisers/co-authorsAcademic job search starts 1 year in advance & takes hoursper application - plan accordinglyEmbrace the nomadic lifestyle!

Publishing:

Quality > quantityAvoid �measuring contests�: # of pubs, impact factors,citationsThere are many good journals: your work doesn't have to bein Science to be impactfulPublishing in Statistics journals take a long time (my Stat inMed paper was submitted in March)

General advice

Graduate school and academic lifestyle:

Learn to fail �better�, learn to �get over� failures quicklyStand up for your research: argue (respectfully) with yourmentors/advisers/co-authorsAcademic job search starts 1 year in advance & takes hoursper application - plan accordinglyEmbrace the nomadic lifestyle!

Publishing:

Quality > quantityAvoid �measuring contests�: # of pubs, impact factors,citationsThere are many good journals: your work doesn't have to bein Science to be impactfulPublishing in Statistics journals take a long time (my Stat inMed paper was submitted in March)

General advice

Working with data:

Back-up everything! Use Dropbox (automatic roll-back toprevious versions)Save R workspaces every so oftenMake sure all IRB documents are signed if you use someone'sdata

Make friends:

Other grad students/post-docs know exactly what you aregoing through

General advice

Working with data:

Back-up everything! Use Dropbox (automatic roll-back toprevious versions)Save R workspaces every so oftenMake sure all IRB documents are signed if you use someone'sdata

Make friends:

Other grad students/post-docs know exactly what you aregoing through

Thank you! Questions?

IntroductionMy projectsPost-doc DetailsFrom UNL to Post-doc

anm0:

life as a post-doc at the national cancer institute€¦ · as a post-doc, my main job is to...

Documents